The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →To tailor Claude Code to a project, start with three files: CLAUDE.md for context that is useful across sessions, a skill’s SKILL.md for a reusable procedure, and .claude/settings.json for configuration and hooks. They do not make Claude universally “smarter”; they give it project-specific guidance and let you define repeatable behavior.
What belongs in each of the three files?
| File | Best for | When it takes effect |
|---|---|---|
CLAUDE.md |
Project facts and conventions useful for many tasks | Root guidance is available from session start; subdirectory guidance can apply when working in that area |
A skill’s SKILL.md |
A reusable, task-specific workflow | The full instructions load when the skill is invoked or matched to a task |
.claude/settings.json |
Claude Code configuration, permissions, and hook registration | Settings determine configuration scope; hooks can trigger actions predictably |
These are practical roles, not an Anthropic-prescribed fixed trio. Anthropic’s June 18, 2026 customization guide describes several ways to steer Claude Code, including rules and subagents as well as these mechanisms.
![Claude Code for Beginners: [All in 1] The Ultimate Step-by-Step Guide to Set Up Safely, Build Real Projects, and Automate Your Work with AI – No Tech Background and No Terminal Experience Needed](https://m.media-amazon.com/images/I/41IO6Hnt6vL._SL160_.jpg)
As an Amazon Associate I earn from qualifying purchases.
What should you put in CLAUDE.md?
Use CLAUDE.md as a concise project briefing: information Claude should have for a broad range of work in that project. Anthropic’s Help Center guidance, dated April 15, 2026, says a Markdown file is automatically read at the start of a session in its directory. Examples include a global ~/.claude/CLAUDE.md and a repository-root file.
Useful project-wide context
- Build, test, and lint commands.
- A brief directory map and architecture overview.
- Coding conventions and team norms that apply across much of the repository.
Keep it focused. Anthropic’s June 18, 2026 article recommends keeping CLAUDE.md under 200 lines, assigning it an owner, and reviewing changes like code. That is a recommendation, not a product-enforced limit. Always-loaded guidance uses context even when a particular task does not need it, so move file-specific conventions into path-scoped rules or subdirectory guidance instead of accumulating them in the root briefing.
When should you create a skill’s SKILL.md?
Put a repeatable procedure in a skill when it is useful for particular tasks but does not belong in every session’s project briefing. Examples include a release checklist or a code-review workflow. A skill can hold the detailed steps and supporting material while CLAUDE.md stays short. The skill name and description can be available at session start; its full instructions load when Claude invokes or matches the skill to the task.
Where do Claude Code hooks go?
Hooks are registered through settings, including project settings such as .claude/settings.json. Depending on the intended scope, configuration can also live in user settings, managed policy settings, or skill or agent frontmatter. Anthropic’s hook configuration guide describes project, user, and local settings scopes: project settings can be shared with a repository, user settings apply across projects, and local project settings are for personal configuration that is not shared.
Use hooks for reliable triggers
Choose a hook when an action should happen predictably, such as running a linter after edits or blocking a command before execution. Command, HTTP, and MCP tool hooks can perform deterministic actions; prompt and agent hooks rely on Claude’s judgment, as Anthropic explains in its customization guide.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Handle hook scripts carefully
Anthropic advises validating and sanitizing hook input, quoting shell variables, using absolute paths, and avoiding sensitive files such as .env and credentials. Treat a hook as executable automation: define its scope and inputs deliberately, rather than assuming that placing it in settings makes it harmless.
Should a constraint go in a skill or a path-scoped rule?
Use a path-scoped rules file when a constraint applies only to matching files—for example, a convention specific to one directory or file type. A skill is better for a workflow Claude should follow when performing a particular kind of task. The distinction is scope: a rule constrains work in its matching paths, while a skill packages a procedure for relevant tasks.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How do you choose the right file?
- Most sessions, most of the project: put durable project or team context in
CLAUDE.md. - Some tasks, repeatable procedure: put the workflow in a skill’s
SKILL.md. - Only particular paths: use path-scoped rules or relevant subdirectory guidance.
- Reliable trigger and action: configure a hook in settings, choosing the scope that fits who should use it.
Anthropic’s Claude Code best-practices article is another primary-source reference on working with Claude Code. The mechanisms above tailor its context and behavior; the sources do not establish a measured, universal improvement in code quality or productivity.
Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
